The AI hype is huge – just ask Nvidia – but businesses should start small and make it real

The rise of AI may feel like almost an overnight business disruption… but let’s not forget it has been more than 60 years in the making. 

Such is the power of the hype cycle that we have witnessed since the sensation of ChatGPT last year, you could be forgiven for thinking that the force with which this crashed into boardroom consciousness has rendered strategies for 2024 obsolete without this “new” technology at their core.

As with all hyperbole, the truth is markedly different. While generative AI delivered a publicly tangible expression of our future, perhaps the bigger impact has been how rapidly it has raised the profile of traditional AI.

Much of the discussion since has actually centred on traditional AI applications and optimisations, not the headline-grabbing generative strain. It’s a new conversation about old things – which arguably could deliver more value – but this has made them more real, which is undoubtedly positive progress.

Now, we seem to be in something of a lull after the peak of the hype and early adoption cycle, before the mainstream adoption phase fully kicks in. 

These waves are typical with all new technology. Post-hype, people try to rapidly reach the advertised panacea, only to find out it’s much harder than they thought. Applications eventually emerge that are not as transformational as what was promised, but useful nonetheless. 

Despite this lull, business leaders still feel the pressure to be seen to be advancing or remaining relevant in this space – AI is the “new digital” in terms of a topic that investors want to hear spoken about. I appreciate its importance from that point of view, but this may simply result in “things” being done that are not particularly well thought through. 

However, a handful of use cases will emerge as long-term value creators, which I would expect to centre around content creation and service interfaces that people will go on to adopt in time. Private Equity also has another lever to pull in terms of value creation for portfolio companies – a lever that was actually available some years ago, activated now because the question of “what shall we do with AI” was not being asked in earnest prior to GPT.

In my view, leadership teams need to hold their nerve and focus on educating themselves. If investing budget, I’d advise focusing this on developing prototypes using owned data sets, and without the expectation of creating transformational value, more simply that they learn. 

There will be many less considered approaches that only serve to deliver expensive mistakes in the coming months. The aftermath of this will be the time to benefit from a much greater understanding of what really works, and how to deploy it in the most practical, cost-effective manner. The digital ecosystem now – with or without AI – allows teams to perfect small implementations and then prepare to scale aggressively. Parking the peer pressure and resisting the FOMO will pay off in the long run.

Nvidia – an early winner in the world of AI

A clear winner to emerge from the AI boom to date is undoubtedly Nvidia, demonstrated by its recent stock surge since the turn of the year. Processing power is critical for AI model training (albeit not query execution), and the company’s scale and established infrastructure is arguably decades ahead of its closest competitors and extremely hard to replicate, creating a huge distortion in this market.

Of course, this scenario begs the question of how long AI model training will run for across global enterprises. I expect it to continue for the long term, but we will probably see an unnatural peak as companies build models for the first time before demand eventually levels out. 

This will no doubt be an unpredictable period for businesses as they find their way in this burgeoning AI-enabled world. I find it helpful to maintain a healthy sense of scepticism as we move through the early phases of any new technology. Personal computers promised paperless offices, while ERPs and CRMs promised revolutionary levels of integration and efficiency. In the early stages of AI, too, a fraction of what has been promised will likely be realised, and expectations therefore should be tempered accordingly.

Let’s also remember that human fallibility lies at the root of most of what will go wrong: Hubris, a herd mentality, selective use of facts, or an inability to manage complexity; we must remain acutely aware of this in everything we do in technology. Despite going against much of what’s in our human make-up, prioritising small, real successes in the short term will move the dial much more than expensive, high-profile disasters driven by the hype.